ZHIXIN’s work stage 2

In stage 2, I am going to make the faceted plot more clear than stage 1.

I will divide the region in London into 4 big regions: from city center to further places, see the picture below.

Red represents group “Central”

Blue represents group “Middle”

Red represents group “Outer”

Unmarked region represents group “Further outer”

Original picture from https://www.doogal.co.uk/london_postcodes.php, accessed on nov.25th 2021.

library(tidyverse)
library(tidyr)
library(purrr)
library(scales)
library(broom)

London <- read_csv("../data/London.csv")

posremo <- function(postcode) {
  temp = 0
  for(index in c( 1 : (str_length(postcode)))){
    i = substr(postcode, index, index)
    if(!is.na(as.numeric(i))){
      temp = temp+1
    }
    if(is.na(as.numeric(i)) & temp > 0){
      postcode = str_remove(postcode, i)
    }
  }
  return(postcode)
}

London <- London%>%
  filter(`City/County` == "London")%>%
  separate(`Postal Code` , c("region","subregion"), " ")%>%
  mutate(
    region = map_chr(region, posremo)
  )
# Until now, this is the same code in stage 1, but now I'm doing a further step - dividing them by large groups

central <- c("WC","EC","N1","E1","E2","SE1","SW1","W1","NW1")

subcentral <- c("N6","N19","N7","N4","N5","N16","E5","E8","E9","E3","E14","SE16","SE8","SE14","SE4","SE17","SE15","SE5","SE11","SW4","SW8","SW9","SW11","SW3","SW10","SW5","SW7","W2","W8","W9","W11","NW3","NW5","NW6","NW8")

suburban <- c("N3","N2","N10","N22","N8","N17","N15","E17","E10","E11","E20","E15","E7","E12","E6","E13","E16","SE10","SE7","SE3","SE13","SE6","SE23","SE22","SE24","SE21","SE26","SE27","SW2","SW16","SW12","SW17","SW18","SW6","W6","W14","W12","W10","NW10","NW2","NW11")

London <- London%>%
  mutate(
    large_region = case_when(
      region %in% central ~ "central",
      region %in% subcentral ~ "Middle",
      region %in% suburban ~ "Outer",
      !(region %in% central | region %in% subcentral | region %in% suburban) ~ "Further Outer"
    )
  )

Now let’s generate the plot with new groups

# This code chunk may takes longer than others. Still finding more efficient ways to do it
pound_in_M <- label_dollar(
  prefix = "£", 
  suffix = "M",
  big.mark = ".",
  decimal.mark = ","
)
London%>%
  mutate(
    Price_in_M = Price/1000000
  )%>%
  ggplot(aes(x = `Area in sq ft`, y = Price_in_M, color = region))+
  geom_point()+
  facet_wrap(~ large_region)+
  scale_y_continuous(labels = pound_in_M)+
  geom_smooth(method = "lm", se = FALSE)

## `geom_smooth()` using formula 'y ~ x'

London <- London%>%
  mutate(
    price_per_sq_ft = Price/`Area in sq ft`
  )

fit_central <- lm(
  formula = price_per_sq_ft ~ large_region,
  data = London
  )
tidy(fit_central) 

## # A tibble: 4 × 5
##   term                      estimate std.error statistic  p.value
##   <chr>                        <dbl>     <dbl>     <dbl>    <dbl>
## 1 (Intercept)                  1487.      24.0     61.9  0       
## 2 large_regionFurther Outer    -507.      32.0    -15.8  3.18e-54
## 3 large_regionMiddle           -275.      29.1     -9.45 6.77e-21
## 4 large_regionOuter            -620.      32.2    -19.2  9.33e-78

fit_central_fixed <- lm(
  formula = price_per_sq_ft ~ large_region,
  data = London%>%filter(`Area in sq ft` < 14000)
  )
tidy(fit_central_fixed) 

## # A tibble: 4 × 5
##   term                      estimate std.error statistic  p.value
##   <chr>                        <dbl>     <dbl>     <dbl>    <dbl>
## 1 (Intercept)                  1487.      24.0     61.9  0       
## 2 large_regionFurther Outer    -507.      32.0    -15.8  3.18e-54
## 3 large_regionMiddle           -275.      29.1     -9.45 6.77e-21
## 4 large_regionOuter            -621.      32.2    -19.3  4.90e-78

London %>%
  group_by(region) %>%
  summarise(
    price_per_sq_ft = median(price_per_sq_ft),
    large_region = large_region
  ) %>%
  ggplot(aes(x = price_per_sq_ft, fill = large_region)) +
  geom_histogram(binwidth = 40) +
  scale_color_manual(values = c("yellow","#20C5D4","#00FF67","#290916"))

## `summarise()` has grouped output by 'region'. You can override using the `.groups` argument.